fsck: fix error in reporting row number for timestamp errors

Since we process in chunks, we need to add "start" to the row number;
however, we don't need to use this when accessing data['timestamp']
(aka ts)

nilmdb/fsck/fsck.py

@@ -248,23 +248,22 @@ class Fsck(object):
 
                     # Check that we can open bulkdata
                     tab = nilmdb.server.bulkdata.Table(bulk)
-                except FsckFormatError:
+                except FsckFormatError as e:
                     # If there are no files except _format, try deleting
                     # the entire stream; this may remove metadata, but
                     # it's probably unimportant.
                     files = list(os.listdir(bulk))
                     if len(files) > 1:
-                        raise
-                    if len(files) == 1 and files[0] != b'_format':
-                        raise
+                        raise FsckFormatError(f"{path}: can't load _format, "
+                                              f"but data is also present")
 
                     # Since the stream was empty, just remove it
                     self.fix_remove_stream(sid, path, bulk,
                                            "empty, with corrupted format file")
-
-                except FsckError:
-                    raise
-                except Exception as e:
+                except FsckError as e:
+                    raise e
+                except Exception as e: # pragma: no cover
+                    # No coverage because this is an unknown/unexpected error
                     raise FsckError("%s: can't open bulkdata: %s",
                                     path, str(e))
                 tab.close()
@@ -278,6 +277,7 @@ class Fsck(object):
                 fmt = pickle.load(f)
         except Exception as e:
             raise FsckFormatError(f"{path}: can't load _format file ({e})")
+
         if fmt["version"] != 3:
             raise FsckFormatError("%s: bad or unsupported bulkdata version %d",
                                   path, fmt["version"])
@@ -511,18 +511,20 @@ class Fsck(object):
                 # Verify that all timestamps are in range.
                 match = (ts < stime) | (ts >= etime)
                 if match.any():
-                    row = start + numpy.argmax(match)
+                    row = numpy.argmax(match)
                     raise FsckError("%s: data timestamp %d at row %d "
                                     "outside interval range [%d,%d)",
-                                    path, data['timestamp'][row], row,
+                                    path, ts[row], row + start,
                                     stime, etime)
 
                 # Verify that timestamps are monotonic
                 match = numpy.diff(ts) <= 0
                 if match.any():
-                    row = start + numpy.argmax(match)
-                    raise FsckError("%s: non-monotonic timestamp (%d -> %d) "
-                                    "at row %d", path, ts[row], ts[row+1], row)
+                    row = numpy.argmax(match)
+                    raise FsckError("%s: non-monotonic timestamp (%d -> %d)"
+                                    " at row %d", path, ts[row], ts[row+1],
+                                    row + start)
+
                 first_ts = ts[0]
                 if last_ts is not None and first_ts <= last_ts:
                     raise FsckError("%s: first interval timestamp %d is not "

tests/test_fsck.py

@@ -163,3 +163,8 @@ class TestFsck(object):
             raise Exception("hi")
         with assert_raises(Exception):
             foo()
+
+        self.failmsg("test2v", "can't load _format, but data is also present")
+        self.failmsg("test2v1", "bad bulkdata table")
+        self.failmsg("test2v2", "empty, with corrupted format file", fix=False)
+        self.okmsg("test2v2", "empty, with corrupted format file")